CFB amplifiers as comparators, advice?

[Boscoe]

Hi all,

I'm trying to use an EL5166 as a CCD clock driver running at 28MHz which needs a voltage swing from -4.5 to 1.5V and a rise time of 3ns or less. I selected this op amp for it's 6000V/us slew rate and high output current which are the limits I've hit before. However I get a small 200mVpp highly distorted output waveform at any frequency by holding the inverting input at 1.5V and applying a 3.3V square wave signal from the FPGA into the non inverting input. I'm using a +/-4V supply for this test.

I'd appreciate some advice for driving CCD clocks. Here are the requirements:

- High slew rate needs 8V in 4ns rise and fall.
- Maximum 10V working voltage.
- Current output of at least 100mA peak.

This means a lot of high speed voltage comparators are out due to working at a maximum of 5.5V, and MOSFET gate drivers are too slow. Buffered op amps are difficult to get the speed and limit cross conduction.

Thanks!

[nctnico]

Start by not overdriving the input of the opamp. In other words: set the amplification of the OPAMP and feed it a signal with the appropriate input level. Rule number one of using opamps: never use an opamp as a comparator!

[David Hess]

I looked up an old Tektronix CCD design (7D20) and they used a complementary pair of RF bipolar transistors in an open loop class-b common emitter amplifier to do the level shift from logic levels.

[Boscoe]

Start by not overdriving the input of the opamp. In other words: set the amplification of the OPAMP and feed it a signal with the appropriate input level. Rule number one of using opamps: never use an opamp as a comparator!

I've never heard of this. Why is this the case?

[Boscoe]

I looked up an old Tektronix CCD design (7D20) and they used a complementary pair of RF bipolar transistors in an open loop class-b common emitter amplifier to do the level shift from logic levels.

Thanks, I'll get on the simulator and try this out.

[David Hess]

Start by not overdriving the input of the opamp. In other words: set the amplification of the OPAMP and feed it a signal with the appropriate input level. Rule number one of using opamps: never use an opamp as a comparator!

I've never heard of this. Why is this the case?

Operational amplifiers just make poor comparators.  They are not designed to operate with their internal nodes in saturation and frequency compensation limits their speed.

I looked up an old Tektronix CCD design (7D20) and they used a complementary pair of RF bipolar transistors in an open loop class-b common emitter amplifier to do the level shift from logic levels.

Thanks, I'll get on the simulator and try this out.

I would not necessary copy what Tektronix did exactly.  I linked it just to give an idea of what has been done in the past and I think a saturated level shifter is the way to go even if that means a discrete output stage.

A linear amplifier with the output characteristics you specified is feasible but if it also has to operate in saturation, then the linear operation part might as well be skipped.